Sergei Shtylyov [Sat, 5 May 2007 20:03:49 +0000 (22:03 +0200)]
cmd64x: fix multiword and remove single-word DMA support
Fix the multiword DMA and drop the single-word DMA support (which nobody will
miss, I think). In order to do it, a number of changes was necessary:
- rename program_drive_counts() to program_cycle_times(), pass to it cycle's
total/active times instead of the clock counts, and convert them into the
active/recovery clocks there instead of cmd64x_tune_pio() -- this causes
quantize_timing() to also move;
- contrarywise, move all the code handling the address setup timing into
cmd64x_tune_pio(), so that setting MWDMA mode wouldn't change address setup;
- remove from the speedproc() method the bogus code pretending to set the DMA
timings by twiddling bits in the BMIDE status register, handle setting MWDMA
by just calling program_cycle_times(); while at it, improve the style of that
whole switch statement;
- stop fiddling with the DMA capable bits in the speedproc() method -- they do
not enable DMA, and are properly dealt with by the dma_host_{on,off} methods;
- don't set hwif->swdma_mask in the init_hwif() method anymore.
In addition to those changes, do the following:
- in cmd64x_tune_pio(), when writing to ARTTIM23 register preserve the interrupt
status bit, eliminate local_irq_{save|restore}() around this code as there's
*no* actual race with the interrupt handler, and move cmdprintk() to a more
fitting place -- after ide_get_best_pio_mode() call;
- make {arttim|drwtim}_regs arrays single-dimensional, indexed with drive->dn;
- rename {setup|recovery}_counts[] into more fitting {setup|recovery}_values[];
- in the speedproc() method, get rid of the duplicate reads/writes from/to the
UDIDETCRx registers and of the extra variable used to store the transfer mode
value after filtering, use another method of determining master/slave drive,
and cleanup useless parens;
- beautify cmdprintk() output here and there.
While at it, remove meaningless comment about the driver being used only on
UltraSPARC and long non-relevant RCS tag. :-)
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Sergei Shtylyov [Sat, 5 May 2007 20:03:49 +0000 (22:03 +0200)]
sl82c105: DMA support code cleanup (take 4)
Fold the now equivalent code in the ide_dma_check() method into a mere call to
ide_use_dma(). Make config_for_dma() return non-zero if DMA mode has been set
and call it from the ide_dma_check() method instead of ide_dma_on().
Defer writing the DMA timings to the chip registers until DMA is really turned
on (and do not enable IORDY for DMA).
Remove unneeded code from the init_hwif() method, improve its overall looks.
Rename the dma_start(), ide_dma_check(), and ide_dma_lostirq() methods, and
also use more proper hwif->dma_command, fix printk() and comment in the latter
one as well. While at it, cleanup style in several places.
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Sergei Shtylyov [Sat, 5 May 2007 20:03:49 +0000 (22:03 +0200)]
sl82c105: rework PIO support (take 2)
Get rid of the 'pio_speed' member of 'ide_drive_t' that was only used by this
driver by storing the PIO mode timings in the 'drive_data' instead -- this
allows us to greatly simplify the process of "reloading" of the chip's timing
register and do it right in sl82c150_dma_off_quietly() and to get rid of two
extra arguments to config_for_pio() -- which got renamed to sl82c105_tune_pio()
and now returns a PIO mode selected, with ide_config_drive_speed() call moved
into the tuneproc() method, now called sl82c105_tune_drive() with the code to
set drive's 'io_32bit' and 'unmask' flags in its turn moved to its proper place
in the init_hwif() method.
Also, while at it, rename get_timing_sl82c105() into get_pio_timings() and get
rid of the code in it clamping cycle counts to 32 which was both incorrect and
never executed anyway...
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Linus Torvalds [Sat, 5 May 2007 04:44:34 +0000 (21:44 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc: (46 commits)
mmc-omap: Clean up omap set_ios and make MMC_POWER_ON work
mmc-omap: Fix omap to use MMC_POWER_ON
mmc-omap: add missing '\n'
mmc: make tifm_sd_set_dma_data() static
mmc: remove old card states
mmc: support unsafe resume of cards
mmc: separate out reading EXT_CSD
mmc: break apart switch function
MMC: Fix handling of low-voltage cards
MMC: Consolidate voltage definitions
mmc: add bus handler
wbsd: check for data opcode earlier
mmc: Separate out protocol ops
mmc: Move core functions to subdir
mmc: deprecate mmc bus topology
mmc: remove card upon suspend
mmc: allow suspended block driver to be removed
mmc: Flush pending detects on host removal
mmc: Move host and card drivers to subdirs
mmc: Move queue functions to mmc_block
...
Linus Torvalds [Sat, 5 May 2007 03:44:54 +0000 (20:44 -0700)]
Merge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2
* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2:
ocfs2: Force use of GFP_NOFS in ocfs2_write()
ocfs2: fix sparse warnings in fs/ocfs2/cluster
ocfs2: fix sparse warnings in fs/ocfs2/dlm
ocfs2: fix sparse warnings in fs/ocfs2
[PATCH] Copy i_flags to ocfs2 inode flags on write
[PATCH] ocfs2: use __set_current_state()
ocfs2: Wrap access of directory allocations with ip_alloc_sem.
[PATCH] fs/ocfs2/: make 3 functions static
ocfs2: Implement compat_ioctl()
Linus Torvalds [Sat, 5 May 2007 03:44:23 +0000 (20:44 -0700)]
iomap: make the default iomap functions fail softer
We used to BUG_ON() for a badly mapped IO port, which is certainly
correct, but actually made it harder to debug the case where the ATA
drivers had incorrectly mapped a nonconnected ATA port.
So make badly mapped ports trigger a WARN_ON(), and throw the IO away
instead (and return all ones for reads). For things like broken driver
initialization - which is the most likely cause anyway - that should
mean that the machine comes up and is usable (at least that was the case
for the ATA breakage that triggered this patch).
It tends to be a whole lot easier to do a "dmesg" on a working machine
than to try to capture logs off a dead one.
Linus Torvalds [Sat, 5 May 2007 02:55:11 +0000 (19:55 -0700)]
Merge git://git.linux-nfs.org/pub/linux/nfs-2.6
* git://git.linux-nfs.org/pub/linux/nfs-2.6: (28 commits)
NFS: Fix a compile glitch on 64-bit systems
NFS: Clean up nfs_create_request comments
spkm3: initialize hash
spkm3: remove bad kfree, unnecessary export
spkm3: fix spkm3's use of hmac
NFS4: invalidate cached acl on setacl
NFS: Fix directory caching problem - with test case and patch.
NFS: Set meaningful value for fattr->time_start in readdirplus results.
NFS: Added support to turn off the NFSv3 READDIRPLUS RPC.
SUNRPC: RPC client should retry with different versions of rpcbind
SUNRPC: remove old portmapper
NFS: switch NFSROOT to use new rpcbind client
SUNRPC: switch the RPC server to use the new rpcbind registration API
SUNRPC: switch socket-based RPC transports to use rpcbind
SUNRPC: introduce rpcbind: replacement for in-kernel portmapper
SUNRPC: Eliminate side effects from rpc_malloc
SUNRPC: RPC buffer size estimates are too large
NLM: Shrink the maximum request size of NLM4 requests
NFS: Use pgoff_t in structures and functions that pass page cache offsets
NFS: Clean up nfs_sync_mapping_wait()
...
* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6: (49 commits)
[SCTP]: Set assoc_id correctly during INIT collision.
[SCTP]: Re-order SCTP initializations to avoid race with sctp_rcv()
[SCTP]: Fix the SO_REUSEADDR handling to be similar to TCP.
[SCTP]: Verify all destination ports in sctp_connectx.
[XFRM] SPD info TLV aggregation
[XFRM] SAD info TLV aggregationx
[AF_RXRPC]: Sort out MTU handling.
[AF_IUCV/IUCV] : Add missing section annotations
[AF_IUCV]: Implementation of a skb backlog queue
[NETLINK]: Remove bogus BUG_ON
[IPV6]: Some cleanups in include/net/ipv6.h
[TCP]: zero out rx_opt in tcp_disconnect()
[BNX2]: Fix TSO problem with small MSS.
[NET]: Rework dev_base via list_head (v3)
[TCP] Highspeed: Limited slow-start is nowadays in tcp_slow_start
[BNX2]: Update version and reldate.
[BNX2]: Print bus information for PCIE devices.
[BNX2]: Add 1-shot MSI handler for 5709.
[BNX2]: Restructure PHY event handling.
[BNX2]: Add indirect spinlock.
...
Linus Torvalds [Sat, 5 May 2007 01:13:17 +0000 (18:13 -0700)]
Merge branch 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/dtor/input
* 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/dtor/input: (65 commits)
Input: gpio_keys - add support for switches (EV_SW)
Input: cobalt_btns - convert to use polldev library
Input: add skeleton for simple polled devices
Input: update some documentation
Input: wistron - fix typo in keymap for Acer TM610
Input: add input_set_capability() helper
Input: i8042 - add Fujitsu touchscreen/touchpad PNP IDs
Input: i8042 - add Panasonic CF-29 to nomux list
Input: lifebook - split into 2 devices
Input: lifebook - add signature of Panasonic CF-29
Input: lifebook - activate 6-byte protocol on select models
Input: lifebook - work properly on Panasonic CF-18
Input: cobalt buttons - separate device and driver registration
Input: ati_remote - make button repeat sensitivity configurable
Input: pxa27x - do not use deprecated SA_INTERRUPT flag
Input: ucb1400 - make delays configurable
Input: misc devices - switch to using input_dev->dev.parent
Input: joysticks - switch to using input_dev->dev.parent
Input: touchscreens - switch to using input_dev->dev.parent
Input: mice - switch to using input_dev->dev.parent
...
Fixed up conflicts with core device model removal of "struct subsystem" manually.
* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/driver-2.6:
remove "struct subsystem" as it is no longer needed
sysfs: printk format warning
DOC: Fix wrong identifier name in Documentation/driver-model/devres.txt
platform: reorder platform_device_del
Driver core: fix show_uevent from taking up way too much stack
* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/pci-2.6: (59 commits)
PCI: Free resource files in error path of pci_create_sysfs_dev_files()
pci-quirks: disable MSI on RS400-200 and RS480
PCI hotplug: Use menuconfig objects
PCI: ZT5550 CPCI Hotplug driver fix
PCI: rpaphp: Remove semaphores
PCI: rpaphp: Ensure more pcibios_add/pcibios_remove symmetry
PCI: rpaphp: Use pcibios_remove_pci_devices() symmetrically
PCI: rpaphp: Document is_php_dn()
PCI: rpaphp: Document find_php_slot()
PCI: rpaphp: Rename rpaphp_register_pci_slot() to rpaphp_enable_slot()
PCI: rpaphp: refactor tail call to rpaphp_register_slot()
PCI: rpaphp: remove rpaphp_set_attention_status()
PCI: rpaphp: remove print_slot_pci_funcs()
PCI: rpaphp: Remove setup_pci_slot()
PCI: rpaphp: remove a call that does nothing but a pointer lookup
PCI: rpaphp: Remove another wrappered function
PCI: rpaphp: Remve another call that is a wrapper
PCI: rpaphp: remove a function that does nothing but wrap debug printks
PCI: rpaphp: Remove un-needed goto
PCI: rpaphp: Fix a memleak; slot->location string was never freed
...
ps3av:
- Move the definition of struct ps3av to ps3av.c, as it's locally used only.
- Kill ps3av.sem, use the existing ps3av.mutex instead.
- Make the 512-byte buffer in ps3av_do_pkt() static to reduce stack usage.
Its use is protected by a semaphore anyway.
Convert non-highmem kmap_atomic() to static inline function
Convert kmap_atomic() in the non-highmem case from a macro to a static
inline function, for better type-checking and the ability to pass void
pointers instead of struct page pointers.
Finn Thain [Wed, 2 May 2007 02:55:56 +0000 (12:55 +1000)]
SONIC interrupt handling
Install the built-in macsonic interrupt handler on both IRQs when using
via_alt_mapping. Otherwise the rare interrupt that still comes from the
nubus slot will wedge the nubus.
$ cat /proc/interrupts
auto 2: 89176 via2
auto 3: 744367 sonic
auto 4: 0 scc
auto 6: 318363 via1
auto 7: 0 NMI
mac 9: 119413 framebuffer vbl
mac 10: 1971 ADB
mac 14: 198517 timer
mac 17: 89104 nubus
mac 19: 72 Mac ESP SCSI
mac 56: 629 sonic
mac 62: 1142593 ide0
Version 1 of this patch had a bug where a nubus sonic card would register
two interrupt handlers. Only a built-in sonic needs both.
Versions 2 and 3 needed some cleanups, as Raylynn Knight and Christoph
Hellwig pointed out (thanks).
Signed-off-by: Finn Thain <fthain@telegraphics.com.au> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Finn Thain [Tue, 1 May 2007 20:33:01 +0000 (22:33 +0200)]
m68k: macmace fixes
Fix a race condition in the transmit code, where the dma interrupt could update
the free tx buffer count concurrently and wedge the tx queue.
Fix the misuse of the rx frame status and rx frame length registers: no more
"fifo overrun" errors caused by the OFLOW bit being tested in the frame length
register (instead of the status register), and no more missed packets due to
incorrect length taken from status register (instead of the frame length
register).
Fix a panic (skb_over_panic BUG) caused by allocating and then copying an
incoming packet while the packet length register was changing.
Cut-and-paste the reset code from the powermac mace driver (mace.c), so the NIC
functions when MacOS does not initialise it (important for anyone wanting to
use the Emile boot loader).
Cut-and-paste the error counting and timeout recovery code from mace.c.
Fix over allocation of rx buffer memory (it's page order, not page count).
Converted to driver model.
Converted to DMA API.
Since I've run out of ways to make it fail, and since it performs well now,
promote the driver from EXPERIMENTAL status. Tested on both quadra 840av and
660av.
Finn Thain [Tue, 1 May 2007 20:33:00 +0000 (22:33 +0200)]
CUDA ADB fixes
Fix the flakiness in the CUDA ADB driver on m68k macs (keypresses getting
wedged down or ADB just going AWOL altogether).
The only IRQ used by this driver is the VIA shift register IRQ. The PowerMac
conditional code disables the other VIA IRQ sources, so don't mess with the
other IRQ flags in the common code -- m68k macs need them.
When polling, don't disable local interrupts when we only need to disable the
CUDA interrupt.
Unless polling, don't clear the shift register IRQ flag. On m68k macs this
creates a race that often breaks CUDA ADB.
Tested on Quadra 840av and LC630 (both m68k); also Beige G3 (powerpc).
Finn Thain [Tue, 1 May 2007 20:32:59 +0000 (22:32 +0200)]
m68k: Mac II ADB fixes
Fix a crash caused by requests placed in the queue with the completed flag
already set. This lead to some ADB_SYNC requests returning early and their
request structs being popped off the stack while still queued. Stack corruption
ensued or an invalid request callback pointer was invoked or both. Eliminate
macii_retransmit() and its buggy implementation of macii_write(). Have
macii_queue_poll() fully initialise the request queues.
Fix a bug in macii_queue_poll() where the last_req pointer was not being set.
This caused some requests to leave the queue before being completed (and would
also corrupt the stack under certain conditions).
Fix a race in macii_start that could set the state machine to "reading" while
current_req was null.
No longer send poll commands with the ADBREQ_REPLY flag -- doing that caused
the replies to be stored in the request buffer where they were forgotten
about.
Don't autopoll by continuously sending new Talk commands. Get the controller to
do that for us. This reduces the ADB interrupt rate on an idle bus to about 5
per second. Only autopoll the devices that were probed.
Explicitly clear the interrupt flag when polling.
Use disable_irq rather than local_irq_save when polling.
Remove excess local_irq_save/restore pairs.
Improve bus timeout and service request detection.
Finn Thain [Tue, 1 May 2007 20:32:58 +0000 (22:32 +0200)]
m68k: Mac IRQ cleanup
There are no slow IRQs on Macs since Roman Zippel's IRQ reorganisation that
went into 2.6.16 and removed mac_irq_list[] and the do_mac_irq_list()
dispatcher. (They were implemented in do_mac_irq_list() by lowering the IPL.)
Hence there's no more use for mutual exclusion in the Mac interrupt
dispatchers. Remove it.
Finn Thain [Tue, 1 May 2007 20:32:57 +0000 (22:32 +0200)]
m68k: Mac nubus IRQ fixes (plan E)
Some Macs lack a slot interrupt enable register. So the existing code makes
disabled and unregistered slot IRQ lines outputs set high. This seems to work
on quadras, but does not work on genuine VIAs (perhaps the card still succeeds
in pulling the line low, or perhaps because this increases the settle time on
the port A input, meaning that the CA1 IRQ could fire before the slot line
reads active).
Because of this, the nubus_active flags were used to mask IRQs, which is
actually worse than the problem it tries to solve. Any interrupt masked by
nubus_active will remain asserted and prevent further transitions on CA1. And
so the nubus gets wedged regardless of hardware (emulated VIA ASIC, real VIA
chip or RBV).
The best solution to this hardware limitation of genuine VIAs is to disable the
umbrella SLOTS IRQ when disabling a slot on those machines. Unfortunately, this
means all slot IRQs get disabled when any slot IRQ is disabled. But it is only
a problem when there's more than 1 device using nubus interrupts.
Another potential problem for genuine VIAs is an unregistered nubus IRQ.
Eventually it will be possible to enable the CA1 interrupt by installing its
handler only _after_ all nubus drivers have loaded but _before_ the kernel
needs them, at which time this last problem can be fixed. For now it can be
worked around:
- disable MacOS extensions
- don't boot MacOS (use the Emile bootloader instead)
- get the bootloaders to disable ROM drivers (Penguin does this for video
cards already, don't know about Emile)
- physically remove unsupported cards
Finn Thain [Tue, 1 May 2007 20:32:56 +0000 (22:32 +0200)]
m68k: Mac IRQ prep
Make sure that there are no slot IRQs asserted before leaving the nubus
handler. If there are and we don't then the nubus gets wedged because this
prevents a CA1 transition, which means no more nubus IRQs.
Make the interrupt dispatch loops terminate sooner.
Explicitly initialise the VIA latches to make the code more easily understood.
Macintosh CS89x0 Ethernet: Netif updates
Addition of netif_stop_queue() before transmission by Michael Schmitz
skb_copy_{from,to}_linear_data() conversion by Geert Uytterhoeven
Michael Schmitz [Tue, 1 May 2007 20:32:39 +0000 (22:32 +0200)]
m68k: Atari fb revival
Update the atari fb to 2.6 by Michael Schmitz,
Reformatting and rewrite of bit plane functions by Roman Zippel,
A few more fixes by Geert Uytterhoeven.
Signed-off-by: Michael Schmitz <schmitz@debian.org> Signed-off-by: Roman Zippel <zippel@linux-m68k.org> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Michael Schmitz [Tue, 1 May 2007 20:32:35 +0000 (22:32 +0200)]
m68k: Atari SCSI revival
SCSI should be working on a TT (but someone should really try!) but causes
trouble on a Falcon (as in: it ate a filesystem of mine) at least when
used concurrently with IDE. I have the notion it's because locking of the
ST-DMA interrupt by IDE is broken in 2.6 (the IDE driver always complains
about trying to release an already-released ST-DMA). Needs more work, but
that's on the IDE or m68k interrupt side rather than SCSI.
Signed-off-by: Michael Schmitz <schmitz@debian.org> Signed-off-by: Roman Zippel <zippel@linux-m68k.org> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Sat, 5 May 2007 00:46:27 +0000 (17:46 -0700)]
Merge branch 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6
* 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6: (44 commits)
i2c-s3c2410: Fix bug in releasing driver
i2c-s3c2410: Fix I2C SDA to SCL setup time
i2c: New i2c-tiny-usb bus driver
i2c: Documentation update
i2c: SPIN_LOCK_UNLOCKED cleanup
i2c: Obsolete i2c-ixp2000, i2c-ixp4xx and scx200_i2c
i2c: New Simtec I2C bus driver
i2c: Bitbanging I2C bus driver using the GPIO API
Use menuconfig objects - I2C
i2c: Restore i2c_smbus_read_block_data
i2c-pxa: Clean transaction stop
i2c-algo-bit: Improve debugging
i2c-algo-bit: Implement a 50/50 SCL duty cycle
i2c-omap: Switch to static adapter numbering
i2c: Blackfin Two Wire Interface driver
i2c-algo-sgi: Comment and whitespace cleanups
i2c: Make i2c_del_driver a void function
i2c: Move i2c-isa-only exported symbol declarations
i2c: Document i2c_new_device()
i2c: Add i2c_new_probed_device()
...
Fixed trivial conflict in Documentation/feature-removal-schedule.txt manually.
Vlad Yasevich [Fri, 4 May 2007 20:55:27 +0000 (13:55 -0700)]
[SCTP]: Set assoc_id correctly during INIT collision.
During the INIT/COOKIE-ACK collision cases, it's possible to get
into a situation where the association id is not yet set at the time
of the user event generation. As a result, user events have an
association id set to 0 which will confuse applications.
This happens if we hit case B of duplicate cookie processing.
In the particular example found and provided by Oscar Isaula
<Oscar.Isaula@motorola.com>, flow looks like this:
A B
---- INIT-------> (lost)
<---------INIT------
---- INIT-ACK--->
<------ Cookie ECHO
When the Cookie Echo is received, we end up trying to update the
association that was created on A as a result of the (lost) INIT,
but that association doesn't have the ID set yet.
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
[SCTP]: Re-order SCTP initializations to avoid race with sctp_rcv()
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com> Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Vlad Yasevich [Fri, 4 May 2007 20:34:49 +0000 (13:34 -0700)]
[SCTP]: Fix the SO_REUSEADDR handling to be similar to TCP.
Update the SO_REUSEADDR handling to also check for listen state. This
was muliple listening server sockets can't be created and they will
not steal packets from each other.
Reported by Paolo Galtieri <pgaltieri@mvista.com>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Vlad Yasevich [Fri, 4 May 2007 20:34:09 +0000 (13:34 -0700)]
[SCTP]: Verify all destination ports in sctp_connectx.
We need to make sure that all destination ports are the same, since
the association really must not connect to multiple different ports
at once. This was reported on the sctp-impl list.
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
David Howells [Fri, 4 May 2007 19:41:11 +0000 (12:41 -0700)]
[AF_RXRPC]: Sort out MTU handling.
Sort out the MTU determination and handling in AF_RXRPC:
(1) If it's present, parse the additional information supplied by the peer at
the end of the ACK packet (struct ackinfo) to determine the MTU sizes
that peer is willing to support.
(2) Initialise the MTU size to that peer from the kernel's routing records.
(3) Send ACKs rather than ACKALLs as the former carry the additional info,
and the latter do not.
(4) Declare the interface MTU size in outgoing ACKs as a maximum amount of
data that can be stuffed into an RxRPC packet without it having to be
fragmented to come in this computer's NIC.
(5) If sendmsg() is given MSG_MORE then it should allocate an skb of the
maximum size rather than one just big enough for the data it's got left
to process on the theory that there is more data to come that it can
append to that packet.
This means, for example, that if AFS does a large StoreData op, all the
packets barring the last will be filled to the maximum unfragmented size.
Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Jennifer Hunt [Fri, 4 May 2007 19:22:07 +0000 (12:22 -0700)]
[AF_IUCV]: Implementation of a skb backlog queue
With the inital implementation we missed to implement a skb backlog
queue . The result is that socket receive processing tossed packets.
Since AF_IUCV connections are working synchronously it leads to
connection hangs. Problems with read, close and select also occured.
Using a skb backlog queue is fixing all of these problems .
Signed-off-by: Jennifer Hunt <jenhunt@us.ibm.com> Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Patrick McHardy [Fri, 4 May 2007 19:15:11 +0000 (12:15 -0700)]
[NETLINK]: Remove bogus BUG_ON
Remove bogus BUG_ON(mutex_is_locked(nlk_sk(sk)->cb_mutex)), when the
netlink_kernel_create caller specifies an external mutex it might
validly be locked.
Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Trond Myklebust [Fri, 4 May 2007 18:44:06 +0000 (14:44 -0400)]
NFS: Fix a compile glitch on 64-bit systems
fs/nfs/pagelist.c:226: error: conflicting types for 'nfs_pageio_init'
include/linux/nfs_page.h:80: error: previous declaration of 'nfs_pageio_init' was here
Cornelia Huck [Fri, 4 May 2007 16:47:51 +0000 (18:47 +0200)]
[S390] dasd: New read device characteristics and read configuration data.
Instead of the deprecated read_dev_chars() and read_conf_data_lpm(),
implement dasd_generic_read_dev_chars() and dasd_eckd_read_conf_lpm().
These should even recover better from error than the original cio
functions.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cornelia Huck [Fri, 4 May 2007 16:47:50 +0000 (18:47 +0200)]
[S390] cio: Deprecate read_dev_chars() and read_conf_data{,_lpm}().
These helper functions are a leftover from 2.4 sync I/O and are a
notorious source for bugs. They lead to device driver specific code
creeping into cio, and some issues can't really be fixed at all.
Device drivers can easily implement those functions themselves in a
more robust manner, so let's get rid of them.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Commit c1821c2e9711adc3cd298a16b7237c92a2cee78d introduced the
uaccess structure that is used to select the correct set of user
copy functions for the different execution modes (standard vs.
noexec vs. z9 optimized user copy). The uaccess symbol is exported
with EXPORT_SYMBOL_GPL. This breaks all non-gpl modules that use
user copy. To make them work again change the export to
EXPORT_SYMBOL.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
And here's a port of the powerpc patch to get rid of the notifier
chain completely to s390. It's ontop of Martins patch as that one
is in mainline already.
Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Eric Dumazet [Fri, 4 May 2007 00:39:04 +0000 (17:39 -0700)]
[IPV6]: Some cleanups in include/net/ipv6.h
1) struct ip6_flowlabel : moves 'users' field to avoid two 32bits
holes for 64bit arches. Shrinks by 8 bytes sizeof(struct
ip6_flowlabel)
2) ipv6_addr_cmp() and ipv6_addr_copy() dont need (void *) casts :
Compiler might take into account natural alignement of in6_addr
structs to emit better code for memcpy()/memcmp() Casts to (void *)
force byte accesses.
3) ipv6_addr_prefix() optimization :
Better to clear whole struct, as compiler can emit better code for
memset(addr, 0, 16) (2 stores on x86_64), and avoid some conditional
branches.
# size vmlinux.after vmlinux.before
text data bss dec hex filename 5262262 647612 557432 6467306 62aeea vmlinux.after 5262550 647612 557432 6467594 62b00a vmlinux.before
thats 288 bytes saved.
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Srinivas Aji [Fri, 4 May 2007 00:32:28 +0000 (17:32 -0700)]
[TCP]: zero out rx_opt in tcp_disconnect()
When the server drops its connection, NFS client reconnects using the
same socket after disconnecting. If the new connection's SYN,ACK
doesn't contain the TCP timestamp option and the old connection's did,
tp->tcp_header_len is recomputed assuming no timestamp header but
tp->rx_opt.tstamp_ok remains set. Then tcp_build_and_update_options()
adds in a timestamp option past the end of the allocated TCP header,
overwriting TCP data, or when the data is in skb_shinfo(skb)->frags[],
overwriting skb_shinfo(skb) causing a crash soon after. (The issue was
debugged from such a crash.)
Similarly, wscale_ok and sack_ok also get set based on the SYN,ACK
packet but not reset on disconnect, since they are zeroed out at
initialization. The patch zeroes out the entire tp->rx_opt struct in
tcp_disconnect() to avoid this sort of problem.
Signed-off-by: Srinivas Aji <Aji_Srinivas@emc.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Fri, 4 May 2007 00:23:35 +0000 (17:23 -0700)]
[BNX2]: Fix TSO problem with small MSS.
Remove the check for skb->len greater than MTU when doing TSO. When
the destination has a smaller MSS than the source, a TSO packet may
be smaller than the MTU at the source and we still need to process it
as a TSO packet.
Thanks to Brian Ristuccia <bristuccia@starentnetworks.com> for
reporting the problem.
Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelianov [Thu, 3 May 2007 22:13:45 +0000 (15:13 -0700)]
[NET]: Rework dev_base via list_head (v3)
Cleanup of dev_base list use, with the aim to simplify making device
list per-namespace. In almost every occasion, use of dev_base variable
and dev->next pointer could be easily replaced by for_each_netdev
loop. A few most complicated places were converted to using
first_netdev()/next_netdev().
Signed-off-by: Pavel Emelianov <xemul@openvz.org> Acked-by: Kirill Korotaev <dev@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Thu, 3 May 2007 20:24:48 +0000 (13:24 -0700)]
[BNX2]: Add 1-shot MSI handler for 5709.
The 5709 supports the one-shot MSI handler similar to some of the tg3
chips. In this mode, the MSI disables itself automatically until it
is re-enabled at the end of NAPI poll.
Put the request_irq/free_irq logic in common procedures.
Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Thu, 3 May 2007 20:20:19 +0000 (13:20 -0700)]
[BNX2]: Fix race conditions when calling register_netdev().
Hot-plug scripts can call bnx2_open() as soon as register_netdev() is
called in bnx2_init_one(). We need to call pci_set_drvdata() and
setup everything before calling register_netdev(). netif_carrier_off()
also needs to be moved to bnx2_open() to avoid race conditions with
the irq.
Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Thu, 3 May 2007 20:19:18 +0000 (13:19 -0700)]
[BNX2]: Add 40-bit DMA workaround for 5708.
The internal PCIE-to-PCIX bridge of the 5708 has the same 40-bit DMA
limitation as some of the tg3 chips. Set dma_mask and persistent DMA
mask to 40-bit to workaround.
Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Thu, 3 May 2007 10:35:31 +0000 (03:35 -0700)]
[NETFILTER]: sip: Fix RTP address NAT
I needed to use this recently to talk to a Cisco server. In my case
I only did SNAT while the Cisco server used a different address for
RTP traffic than the one for SIP. I discovered that nf_nat_sip NATed
the RTP address to the SIP one which was unnecessary but OK. However,
in doing so it did not DNAT the destination address on the RTP traffic
to the Cisco back to the original RTP address.
This patch corrects this by noting down the RTP address and using it
when the expectation fires.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / linux-2.6 / log